[ On Tuesday, September 21, 1999 at 18:32:23 (+0200), eric.delcamp@legrand.fr wrote: ]
> Subject: Re: What happened to the lm75(?) driver?
>
> To avoid most of copyrights problem with Greg A. Wood's work, I
> rewrote the driver and give him credits on the header of sources
> files.

Congratulations and thanks!

I guess now I'll need to find another desirable, but not critical;
significant, but not too large; feature to implement as another test
case to the intellectual property morals of TNFi.  (Or is the debate
still raging in the background?)

Meanwhile I'll offer my driver as-is back to the FreeBSD folks I guess
(since that's where it was first implemented! ;-).  Maybe I can even do
the integration back without going insane now -- I think they have
enough bus* stuff to make it not too far of a stretch....

>   1) Hardware management chip like lm75, lm78, became more and more
> popular and some new models appear on market every days ! It will
> become hard to find a generic format to include most of interesting
> features common to all the chips ;-) (I saw some motherboard with
> cascading lm75 !!)

I faced a similar, but somewhat opposite, dilema when I was designing my
driver.  I ended up building something that was very specific to one
chip (with the thought that it wouldn't be too hard to make it slightly
more generic so that it could work with very similar chips) because even
though I had some glimmerings of what other people were doing the only
actual systems I had to test on were FIC's and ASUS' motherboards.

From my surveys of what seems to be happening in the hardware management
frontier it would seem that there are going to be two common approaches
to hardware design.  One will be the custom use of chips like the LM78
and LM79 and very simple use of I2C busses to connect other monitoring
and management devices (including for eg. LM75 sensor chips).  The other
extreme is Intel, HP, NEC, DEl, et al's IPMI (Intelligent Platform
Management Interface) effort with its baseboard management controller.

FreeBSD is already a quite a bit ahead of NetBSD with much more I2C and
SMB support.  My driver (and perhaps yours too) should work well in this
scenario and should be able to work in conjunction with an LM75 driver
as well.

Whether Intel and its buddies will succeed in making IPMI
implementations common enough is yet to be determined (not very many
people other than Intel make server boards that are well enough
documented to be of any use).  Unfortunately the IPMI specification
seems to be covered by a shrink-wrap non-disclosure.  I certainly can't
personally honour it because I downloaded my copy with anonymous FTP
long before I knew it was supposed to be "confidential"!  ;-) Though of
course I won't give away any more copies since they are copyright ;-)
You can get your own at: ftp://download.intel.com/design/servers/ipmi/*.pdf

The entire IPMI scheme would be very cool to have -- though it is very
ambitious and the hardware that implements it will likely be fairly
expensive for some time yet.  Does anyone out there have detailed
hardware manuals for a big Intel, DELL, or HP server that has such
features already built in?  I'd really love to have one of each in my
"lab" so that I could work on supporting them!  ;-)

The next trick is to figure out how to build the really useful software
on top of these hardware management interfaces!  I don't imagine many
*BSD'ers would adore Intel's suggestion of popping up a dialog box to
warn the user when a fan stops spinning....

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>