REVISITING Inconsistent usage of "assertion and negation" in trainingpattern section of SPI-4

Gerry.Houlder at seagate.com Gerry.Houlder at seagate.com
Wed Mar 27 14:57:11 PST 2002


* From the T10 Reflector (t10 at t10.org), posted by:
* Gerry.Houlder at seagate.com
*
Propsal 02-114 rev 0 was tentatively accepted at the March SPI meeting as
the way to resolve this timing issue. Further investigation has shown that
this was the WRONG resolution. Document 02-114 rev. 1 has the new proposed
resolution and some explanation about why the change is needed. The
document will be on the T10 web site shortly, but the text of that proposal
is added below. This issue must be readdressed at the May working group
meeting.

- - - - - - - - - - - -

In section 10.7.4.2.2, After the words "Start of section A:", there is this
example of normal wording for an assertion and negation requirement:
...
8) simultaneously assert and negate REQ, P1, P_CRCA, and DB(15-0) signals
at the negotiated
transfer period 64 times, (e.g., (2 x 6,25 ns) x 64 = 800 ns at fast-160);
...

This example is representative of most phrases that include the words
"assert and negate". This usage of "64 times" describes 64 cycles (64
assertions and 64 negations) of the referenced signals.

Now compare that usage with section 10.7.4.3.3, after the words "SCSI
target ports shall begin pacing transfers by:"
...
1) simultaneously with the assertion of REQ the SCSI target port shall
begin asserting and negating
P1 at twice the negotiated transfer period (e.g., 12,5 ns for fast-160);
2) SCSI target port shall assert and negate P1 at least 8 times (e.g., (2 x
6,25 ns) x 8 = 100 ns at
fast-160); and
....

Note that this wording says "assert and negate ... 8 times" but the 100ns
timing value only allows for 4 assertions and 4 negations of P1 for the
frequency it must operate at. This is inconsistent editorial use of the
"assert and negate x times" phrase.

Rev. 0: Discussion at the SPI working group on 3/12/2002 can to a consensus
that "8 cycles" was the intended requirement, resulting in a 200 ns
preamble. This was suggested to be the correct fix for the wording.

Rev. 1: Later discussion via phone and email revealed that many companies
had designed around the 100 ns preamble and will require hardware change to
meet the 200 ns requirement. Thus the requirement needs to be stated as 4
cycle / 100ns to avoid requiring several companies to redesign their
hardware to meet the requirement. All companies that responded indicated
that their designs will work with the 100 ns minimum preamble and most
indicated their designs will provide at least 125 ns preamble. Therefore
the wording must change to (as shown between the {} markers):

?
2) SCSI target port shall assert and negate P1 at least  { 4 times (e.g.,
(2 x 12,5 ns) x 4 = 100 ns } at fast-160); and
?

*
* 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