One more time now, and may be just may be, we can put this entire
binary-only
UDI debate to bed.
Since it seems people are not reading the URL's which have been provided
I have decided to quote the entire zdnet posting,
IT Week: Intel looks to Linux community for help with UDI
http://www.zdnet.co.uk/news/1998/37/ns-5501.html
I will than take it apart piece by piece.
<Begin Quote>
Most failures on enterprise systems are due to driver problems,
industry representatives heard at the Intel Developer Forum in
Palm Springs Wednesday. "Driver reliability is the smoking gun
keeping Unix on Intel servers out of the enterprise," said Intel's
Justin Rattner during a server technology seminar where the Linux
community was identified as a source of help for the mighty chip
maker.
Rattner was talking during the week that Intel announced its
decision to support Project UDI (Uniform Driver Interface), a
Unix industry initiative to develop a single device driver
implementation that will work across different Unix operating
systems running on Intel platforms. Project UDI is made up of
representatives from Compaq, Hewlett-Packard, IBM, SCO and Sun
as well as Adaptec, Bit3 and Interphase. Intel is to provide
information on interfaces and initiatives as well as engineering
resources to help complete a reference design. The complete
specification is due to be delivered at the next Intel Developer
Forum in February.
<End Quote>
Several points:
1. Project UDI is not "A UNIX industry initiative"
>From the Project UDI Policies and Procedures Document date November 5
1995
I quote:
<Begin Quote>
1. Purpose
The purpose of the Project UDI working group is to define and promote a
Uniform Driver Interface specification, which would allow device drivers
written to this specification to be supported, without change, by
diverse
operation systems and platform vendors.
Participation in the specification process includes a number of OS
vendors
and IHVs, and is open to new participants at any time.
<End Quote>
2. It is is not Intel specific despite what the above URL would lead
you to believe. Please see above quote from the Project UDI document.
<Begin Quote>
"As we move towards IA-64 there's going to be an increasing
number of Unix vendors," said Michael Demshki, Intel content
manager responsible for UDI. "The big issue in making Unix
available on Intel is not just getting the OS working but
getting all the devices working too." Currently there has to
be a device driver for each implementation, but Project UDI
should change this according to Intel. "We saw this as the
most logical place for us to have an impact, and there was no
reason for us to begin a separate effort."
Kevin Quick, chairman of Project UDI, added that the initiative
will let OS vendors work on core issues on OC content rather
than having to concentrate on hardware issues. "Now, companies
such as SCO have artificial restrictions placed on them," he said.
"A vendor may have drivers for most PCI devices, but perhaps not
for Gigabit Ethernet, and so they'll lose the sale - the
customer will have to go elsewhere."
<End Quote>
Several points:
1. As Intel is aware getting the OS running to one thing,
getting all those device drivers for all those peripherals
is something different.
2. Mr. Kevin Quick, InterPhase, clearly states that by not
having a device driver for a particular device a client may
have sales are lost.
3. Both Intel and Mr. Kevin Quick, make it a point to show
that commercial OS vendors have "artificial restrictions" placed
on them. They do not have all the necessary device drivers they
need to compete.
Project UDI and Intels Solution:
<Begin Quote>
However, writing new drivers for the thousands of peripherals on
the market is a daunting task. Hence, Project UDI is hoping the
Linux community will help. Linux will be, said Quick, key to the
adoption of the UDI initiative. A reference platform will be
distibributed as freeware for Linux, and the Project UDI members
will be counting on the Linux community to work on device drivers.
"We have talked to Linus Torvalds (the creator of Linux) and he
was very interested in the idea," Intel's Demshki said.
"The advantage of releasing to the Linux community is that their
work will give Unix OS vendors a basis to work from," Quick added,
though he stressed that the specification will still be tightly
controlled and standards based.
<End Quote>
Several Points:
Question: Who is being asked to undertake the "daunting"
task of writing all these UDI device drivers?
Answer: The Linux Community
Question: Who wants to use those UDI device drivers written
by the Linux Community?
Answer: The commercial OS vendors, and peripheral vendors.
Question: IF the Linux Community were to write all those
UDI device drivers, would the Linux Community have a need
for binary-only UDI device drivers?
Answer: NO
Question: Should the Linux Community be concerned that Project UDI
would control the UDI specification?
Answer: NO. In siting the Project UDI Policies and Procedures Document
dated November 5th, 1995, we have the following:
<Begin Quote>
4. Participation
Participation is open to all parties, regardless of company affiliation.
Voting status is gained by attending three meetings with a three month
period.
( a new participant can vote at the third meeting). A voting participant
must attend no less than one of every three scheduled meetings in order
to
maintain voting status. The intention is to maintain the level of
participation
so that voting participants are up to speed on the issues. Meeting
attendees who
are not voting participants was welcome to participate in all
discussions and
to vote in non-binding advisory votes ("straw polls).
<End Quote>
Basically this means that if 100 separate individuals, thereby
representing 100
different "companies", from the Linux Community were to attend three
scheduled
meeting of Project UDI, at the third meeting I would tend to believe
that the
Linux Community would dominate Project UDI.
So what do we have:
1. Project UDI, SCO, Intel, Hewlett-Packard, IBM, InterPhase, Adaptec,
and
others need the help of the Linux Community.
2. Mr. Kevin Quick, states this very clearly when he stated that the
Linux
Community is "key" to the adoption of UDI.
What should be the "price" for this help from the Linux Community.
Alan Cox provided the basic idea for what that price has to be.
Since several of the participants in Project UDI are also involved in
I2O,
specifically Intel, Hewlett-Packard, and Adaptec are on the Sterring
Committee
of I2O. The price is as follows:
I2O becomes totally open, just like Project UDI, in fact I2O
would use the following to indicate this:
(This is taken straight from the Project UDI Policies and Procedures
Document dated November 5th, 1995, I edited specifically for I2O)
<Begin Quote>
1. Purpose
The purpose of the I2O working group is to define and promote
a technical specification for high-performance I/O systems.
Participation in the specification process includes a number of
OS vendors and IHVs, and is open to new participants at any time.
Intelluctual Property
The I2O specification is intended to be publicly available for
implementation by anyone, whether or not they are participants
in the working group.
The _definition_ of any specification developed by the working
group will be placed in the public domain, not subject to copyright,
patent, or any other intellectual property right, so that any party
may implement or utilize the specification. However, any party may
develop and assert intellectual property rights over a particular
_implementation_ of the interface.
<End Quote>
If the participants in Project UDI & I2O are unwilling to meet this
"price", the Linux Community ignores them.
They need the Linux Community more the Linux Community needs them.
Gregory Maxwell wrote:
>
> On Sun, 20 Sep 1998, Brandon S. Allbery KF8NH wrote:
>
> > In message <199809201119.MAA06952@killala.koala.ie>, Simon Kenyon writes:
> > +-----
> > | one implication of UDI that i think people are missing is that if it takes off
> > | we have a situation where certain internal interfaces would be frozen, with no
> > | realistic chance of them being changed other that by the "standards" people.
> > | this would leave Linux at a severe disadvantage, because one of the beauties
> > | of linux is that it can evolve without asking anyone's permission.
> > +--->8
> >
> > But the commercial Unixes are at the same disadvantage: Sun can change
> > DDI/DKI whenever necessary, etc. UDI should be considered a sort of
> > "baseline", with native driver support used for performance or enhanced
> > functionality, etc.
> >
>
> Lets assume that there will be more closed drivers because of UDI. (most
> people agree on this)..
>
> Because of this it will be harder for us to make native drivers, because
> we wont be able get the specs and companies wont share them with us
> because our drivers are open.. Compaines usually wont make native drivers
> because the UDI will be 'good enough'.
>
> Now, you say that the commercial unixes will be at the same
> disadvantage... Not so. They are happy to release binary only native
> drivers. They can buy the source to the UDI drivers and make good native
> drivers.. We'll be stuck trying to reverse engineer the close UDIs, which
> is not legal everywhere (in the US there are new copyright laws being
> passed that put tough constraints on reverse eng... In other places you
> can only reverse eng. for 'compatiblity' which binary only UDI driver
> arguibly give you..)..
>
-- Terry L. Ridder Blue Danube Software (Blaue Donau Software) "We do not write software, we compose it."entertaining angels by the light of my computer screen 24-7 you wait for me entertaining angels while the night becomes history host of heaven, sing over me ==Entertaining Angels==Newsboys
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/