RE: VPD ECN

[Jeff Dahlin <JDahlin@appiangraphics.com>]

The way I read the VPD ECN, it looks like read/write memory is required:
       ECN page 5, last paragraph:  "There is only one read/write space
required for a board."

BUT I didn't find any descriptions of read/write data that are required.
 So as I see it, to be compatible you must have non-volatile read/write
memory on the board, but you can leave it blank.  Am I missing
something? 

One problem I have with putting an EEPROM on a board is the limited life
of the EEPROM.   It makes the system vulnerable to a destructive virus
that could intentionally write repeatedly to all of the EPROM's on the
PCI bus and wear them out.  If the drivers are looking for specific
data, this could cause a lock-up condition and "ruin" the PCI board.

Still, just requiring a PROM on the board seems too much.  I agree with
Bruce Young that the information will not benefit the main market.  In
fact, the way I read it, the only required data in the ROM are Part
number, FRU Part number, EC level, Manufacture ID and  Serial number.
We currently track all of that information with bar codes and
manufacturing documentation.  The Justification for the ECN says that
the board manufacturer can use the information to identify failures
associated with a point of manufacture or different components.  There
are other methods to do the same thing that most companies already have
in place (e.g. bar codes and manufacturing documentation)   Changing to
use the data in the VPD would probably not be worth the expense.  The
Justification also states that system vendors can use the information
for warrantee tracking and analyzing failures.  This is true, but only
if the system vendor and the board manufacturer agree on a format for
this additional information.  Which can be done under the current spec.

These are nice things to do, but I don't believe that they should be set
forth as requirements by the SIG.

According to the Justification, one of the problems that this ECN was
trying to overcome was that the expansion ROM base address is set to 0
after the configuration is completed and therefore the VPD is not
available after that.  It seems to me that a better solution would be to
require  the BIOS to copy the VPD to system RAM, making it available.
Then the presence of VPD could still be optional.  This would solve the
accessibility problem while impacting fewer products and do it at a
lower overall cost.

The VPD ECR should not be approved.

Jeff Dahlin
Appian Graphics
6640 185th Ave. NE
Redmond, WA  98052
(425) 882-2020
 FAX 882-8618

jdahlin@appiangraphics.com
www.appiangraphics.com



>----------


>From: 	Bruce Young[SMTP:youngbru@pionet.net]
>Sent: 	Thursday, June 26, 1997 11:28 PM
>To: 	Mailing List Recipients
>Subject: 	VPD ECN
>
>Most of you should be aware that there are several ECNs pending in the
>PCI SIG at this time. I would urge all of you to look them over and
>provide comments to the SIG ASAP. The ECNs are available in the members
>only area on the PCI SIG web site at
>http://www.pcisig.com/members/ecn.html. If you do not have access to the
>members only area either contact your company's SIG representative or
>send mail to webmaster@pcisig.com.
>
>I would especially urge you to pay close attention to the VPD ECN. This
>ECN will require every PCI add-in card to add a EEPROM and circuitry to
>read and write it. In my opinion this will add cost to nearly every card
>with very little added benefit to the mainstream marketplace. While the
>information that can be made available in the VPD would be nice to have,
>in my opinion, most of the functionality can be duplicated without
>adding cost to the card. 
>
>Read the ECO and form your own opinion and then send feedback to
>ecn@pcisig.com
>
>
>-Bruce Young
>On the net I speak for myself, not Gateway 2000 (or Intel or anyone
>else)
>
>

Ù0