lkml.org 
[lkml]   [1998]   [Sep]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: Linux, UDI and SCO.
I'd be glad to comment on the issues raised below.  There are some
repeats (both internally and from other postings), but this message
covers some good ground so I've reproduced it intact and commented
throughout.

-Kevin

Terry L Ridder writes:
> Hello;
>
> For background on my comments I would suggest reading the original
> reports at:
>
> IT Week: Intel looks to Linux community for help with UDI
> http://www.zdnet.co.uk/news/1998/37/ns-5501.html
>
> A Brief Quote from the above article is below
>
> <Begin Quote>
> "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>
>
> Uniform Driver Interface (UDI)
> http://www.sco.com/udi/
>
> Below is a brief quote from the above Web Page:
>
> <Begin Quote>
> To demonstrate the feasibility of the UDI architecture and to gain
> real-life experience before finalizing the specification, a prototype
> environment implementation was created and ported to the following
> platforms,
> running a SCSI driver from Adaptec and/or a network interface
> driver from Interphase.
>
> Operating System Processor Type
> Compaq Digital UNIX Alpha (64-bit)
> Hewlett-Packard HP-UX PA-RISC
> IBM AIX PowerPC
> NCR MP-RAS IA-32 (x86)
> SCO OpenServer 5.0.5 IA-32 (x86)
> SCO UnixWare 2.1.3 IA-32 (x86)
> SCO UnixWare 7 IA-32
> (x86)
> Sun Microsystems Solaris Sparc
> <End Quote>
>
> Last but not least read the GNU General Public License at:
>
> GNU General Public License
> http://www.fsf.org/copyleft/gpl.html
>
>
> Notice that the first URL which I quoted clearly indicates that
> the commercial vendors would use the Linux UDI drivers as a basis
> to work from. This is very clearly stated by Quick.
>
> Background on Mr. Quick:
>
> Kevin Quick, chairman of Project UDI.
> Kevin Quick's e-mail address is kquick@iphase.com

More background on me: I manage a team of device driver developers
and am myself a device driver developer. We support a large number of
devices and operating systems, not limited to Unix.

>
> Now notice the proof of concept quote from the Project UDI home page:
>
> Listed here are Digital UNIX, HP-UX, Solaris, AIX, UnixWare 2.1.3,
> UnixWare 7, and OpenServer 5.0.5.
>
> Hardware platforms lists are Alpha, PA-RISC, PowerPC, Intel, and Sparc.
>
> Linux runs on all the above mentioned platforms.
>
> Assume for the moment that the Linux Community does write Linux UDI
> device
> drivers. Let us assume for the moment that all peripheral vendors will
> release
> all needed/wanted/required documentation for their peripherals to the
> Linux
> Community. It would be the option of the UDI device driver author to
> release
> the driver under alternate software licenses in addition to the GNU
> General
> Public License. This would not present a problem for the commercial OS
> vendors
> nor for the peripheral vendors, they would just use the alternate
> software
> license. However, if the author only releases the UDI device driver
> under
> the GNU GPL the commercial OS vendors could not use that driver in
> their own closed source OS. The only way they could use it was if their
> source code was released under the GNU GPL, or *BSD license without the
> advertising clause.

UDI does not stipulate the licensing constraints on device driver
distribution. It is our intent to facilitate both source and
object/executable distributions. The driver author/organization is
free to select any license terms they see fit, including GPL, Public
Domain, NDA, or others.

>
> It is important to note that the commercial OS vendors, and peripheral
> vendors are relying on the Linux Community to perform the "daunting"
> task of writing the UDI device drivers. Below is another brief quote
> from the first URL.

Not necessarily. My point in making the statement was that any OS
supporting UDI could make use of UDI drivers developed by Linux
developers. The converse is also equally valid... that the Linux
community will benefit by "commercially" developed UDI drivers.

Please note that what you read in the press is the editor's impression
of what was said and what was important for his target audience,
limited by number of words or lines for space in that article. I
personally think Matt Loney did a very good job in both his research
(he took considerable time to talk with Intel and myself at the IDF)
and his article. He was not the only editor to do so although it's
unfortunate that there were a number of articles that were not so well
researched or written by other editors.

>
> <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.
> <End Quote>
>
> The first two lines are of special interest to the Linux Community:
>
> "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."

There appear to be a large number of good developers in the Linux
community that have much to contribute to developing solid and fast
device drivers. There are other communities that also have much to
contribute, including commercial and educational institutions.

Don't read too much into the quotes... neither Project UDI nor Intel
is embarking on the enslavement of Linux driver developers to our own ends.

>
> Another important phrase in this quote is made by Mr. Quick:
>
> "Linux will be, said Quick, key to the adoption of the UDI initiative."
>
> Perhaps David would be able to have Mr. Quick clarify his
> statements?

Certainly. Although I think my comments above have done a large part
of this let me amplify somewhat:

* Linux is a strong, viable OS with Unix roots.
* Linux source is available to everyone under GPL.
* Commercial OS vendors compete against each other and Linux on the
basis of many factors and features but also have a large user
community to maintain [not that Linux doesn't, but Linus is not
likely to be fired or sued for violation of a support contract if
"Today Linus redesigns the networking driver interface"].
* Commercial OS vendors need to be very careful when incorporating new
technologies and features into their operating systems to maintain
and support their market position and their customer obligations.
The ability to review a functional implementation of a new
technology in source code format helps them to both understand and
implement that technology.
* Finally, Linux is exerting considerable pressure on Commercial OS
vendors from a competitive basis. *If* UDI becomes a viable reality
for Linux, other OS vendors are that much more likely to adopt it.
The converse is also true: if a commercial OS vendor implements UDI
and benefits from it other OS vendors/implementors are likely to
follow to remain competitive.

>
> Please note that no where in the first URL web page are there any
> statements
> concerning either the commercial OS vendors supporting Linux or the
> peripheral vendors supporting Linux with UDI drivers. The entire point
> of the article is that the Project UDI, the commercial OS vendors, and
> peripheral vendors are "hoping" that the Linux Community takes on this
> "daunting" task.

Neither does the article preclude the above. The article focused on
specific issues in the space that was allowed. Certainly Project UDI
would like to see support from most or all commercial OS vendors and
most or all peripheral vendors (IHV's). You can be sure that the
current list of UDI participants which includes OS vendors and
peripheral vendors are not spending time and resources lightly
irrespective of product-related press releases.

>
> Given that Mr. Quick is clearly indicating that Linux and thereby the
> Linux
> Community are "the key is adoption of the UDI initiative", it would seem
> to
> me that this places the Linux Community in an extremely awkward
> position.
> If we do not support Project UDI, it will be because of "us" that
> Project UDI
> "died on the vine". This would also seem to run the risk of being
> labeled,
> "unsupportive", "you can not count on the Linux Community for support",
> "contrary", etc.

Absolutely not the point. I can see how this is one possible
conclusion that could be drawn, but Project UDI existed prior to the
Linux phenomenon and will continue to go forward irrespecive of the
amount of involvement by the Linux community.

You will not see Project UDI (nor I suspect Intel) blaming the Linux
Community for the success or failure of UDI. Neither are we
attempting to constrain or obligate the Linux Community.

>
> If we do support Project UDI, and the UDI device drivers are only
> released
> under GNU GPL, will we not also be labeled? Yes we supported Project
> UDI,
> but no one other than Linux is able to use the UDI drivers.

I'm not enough of a lawyer to fully respond. My impression is that if
someone releases a UDI driver under GPL that any OS vendor can
redistribute that driver as part of their OS although they are (a) not
allowed to charge for that driver (b) must make the source for
that driver available to their customers upon demand, and (c) if they
modify/enhance the driver and distribute it as part of the OS that
both (a) and (b) apply *and* they must include the modifications as
part of (b).

Please let me know if my above interpretation is correct.

>
> Please also note that neither the Porject UDI nor the ZDnet article Web
> Page
> give any indication that all the peripheral vendors will provide the
> needed/wanted/desired/required technical documentation that would be
> needed
> to write the UDI device drivers. There are two peripheral vendors
> mentioned
> on the Project UDI Web Page namely Adaptec (recently joined Linux
> International),
> and Interphase.

It's possible that UDI would have the adverse effect that HW vendors
would produce a UDI driver and no device documentation but I'd think
this would be minor: presently HW vendors which tend to not provide
specifications already don't provide the specifications and
organizations like Linux are out-of-luck... with UDI at least there's
the possibility that you'll get a driver that works under Linux even
if the HW vendor doesn't publish documentation.

>
> There is also no mention of any of the video chip manufactuerers backing
> Project
> UDI. Having worked on the XFree86 project in the past, we would need the
> support of S3, Cirrus Logic, etc.

Video is one of many technologies not currently addressed in the UDI
specification. Naturally Project UDI would not presume to generate a
standard for video or any other technology in the absence of any
experience or expertise in that technology. We'd love to get more
participants and broaden the coverage for devices and technologies
that much quicker.

>
> Assume the Linux Community does support Project UDI there is still no
> guarantee
> that each and every peripheral marketed will be supported. If some
> commercial
> OS vendor can not make sales because there is no UDI driver for some
> clients
> particular brand/model/make/etc of peripheral card is the Linux
> Community going
> to be blamed for the lack of support?

I think I dealt with most of this above... one other thought
however: if the OS vendor cannot make sales because there's no driver
(UDI or otherwise) they'll most probably do what they already do:
(a) (strongly) encourage the vendor to provide a useable driver
(b) develop a driver internally
(c) choose a competing vendors product that has a useable driver

>
> In some respect Mr. Quick has jumped the gun by making these statements.
> The UDI reference platform is not due out till February 1999 when the
> complete
> specification is released at the next Intel Developer Forum. It is not
> until
> then that the Linux Community particularly Linus sees what changes would
> be
> required to the Linux kernel to accomadate the UDI device drivers.

See above. Also the specs are available (although changing) now.

> Given the awkward position that Mr. Quick and Intel representatives have
> placed the Linux Community by this annoucement, what is going to be the
> reaction of those outside of the Linux Community if Linus decides not
> to accomadate UDI?

Linus controls Linux. UDI will not become part of the Linux core
unless Linus accepts UDI.

Intel can provide UDI for Linux as an add-on even if it isn't accepted
into the Linux core. In this respect its like many other products
which run on Linux.

This does bring up a good point however: UDI does not preclude the
use of existing device drivers architectures. While the Project UDI
folks hope that we are developing something which becomes so
universally accepted that it replaces existing kernel architectures we
don't necessarily have the goal of world domination in mind. We also
recognize that to invalidate the entire body of existing drivers the
moment UDI is introduced into an OS is also not feasible. UDI is
designed to be able to operate side-by-side with existing drivers in
the OS and allows a phased evolution to UDI (or not) as desired.

>
> I am open to suggestions.
>
> I again see only two good solutions to the current awkward position
> Mr. Quick seems to have placed both Project UDI, and the Linux
> Community in.
>
> There are basically only two ways this UDI scenario can work.
>
> 1. All the commercial backers of it, switch to using Linux for their OS
> and they just build hardware, or in SCO's case additional software
> add-ons.
>
> 2. SCO releases all UNIX source code under the GNU GPL,
> HP releases all sources code of HP-UX,
> Sun releases all source code for SunOS and Solaris.
> You might as well ask Apple to release all the source code to
> NextStep/OpenStep/Rhapsody.

I don't think either of these are going to happen. However, I hope my
explanations above have shown why neither of these is necessary.

>
> In either scenario there would be no problem if the UDI drivers were
> released only under the GNU GPL.
>
> Once again I ask who is going to represent the Linux Community in this
> current situation? It is clear that someone preferablly a couple of
> people should represent the Linux Community, and keep the rest of the
> Community informed of the current status of Project UDI, the
> reference platform, and the release of the complete UDI
> specification.
>
> Since the commercial OS vendors, and peripheral vendors are making it
> known here and now that Linux is the key to UDI adoption, and that the
> Linux Community is being asked to help in the "daunting" task of writing
> the UDI drivers so that the commercial OS vendors can use our work as
> a basis, we as a community better have some say in the Project UDI.

Please do. Read the specs and join us in completing the specification
or enhancing it in areas (such as video) which are not yet covered.

>
>
> David Hollister wrote:
> >
> > Terry L Ridder wrote:
> > >
> > > Hello;
> > >
> > > There are basically only two ways this UDI scenario can work.
> > >
> > > 1. All the commercial backers of it, switch to using Linux for their OS
> > > and they just build hardware, or in SCO's case additional software
> > > add-ons.
> >
> > In addition to what I say below, I don't understand what you're getting
> > at by this statement either.
> >
> > > 2. SCO releases all UNIX source code under the GNU GPL, HP releases all
> > > sources code of HP-UX, Sun releases all source code for SunOS and
> > > Solaris.
> > > You might as well ask Apple to release all the source code to
> > > NextStep/OpenStep/
> > > Rhapsody.
> >
> > Why would the OS guys have to release their OS source code? Their
> > source code has nothing to do with a Linux driver written to conform to
> > UDI. The only piece of UDI code that is of any real concern to the
> > Linux community is the Linux OS environment piece. THAT would have to
> > be released under the GPL for it to be publicly accepted. Anybody who
> > wrote UDI drivers for Linux would also want to release their drivers
> > under the GPL. In that case, the entire Linux UDI driver environment is
> > then released GPL. What does HP-UX, SunOS, etc. have to do with
> > anything?
>
> Please see above.
>
> >
> > Maybe I'm misunderstanding your point... Or maybe there is a lack of
> > understanding by many about how UDI is architected.
> >
> > --
> > David Hollister Interphase Corporation dhollist@iphase.com
> > Software Engineer Dallas, TX
>
> --
> Terry L. Ridder
> Blue Danube Software (Blaue Donau Software)
> "We do not write software, we compose it."
>
> When the toast is burnt
> and all the milk has turned
> and Captain Crunch is waving farewell
> when the Big One finds you
> may this song remind you that they
> don't serve breakfast in hell
> ==Breakfast==Newsboys
>
>

Thanks for the opportunity to respond to the above concerns. Feel
free to follow-up with me if you're still concerned.

--
________________________________________________________________________
Kevin Quick Interphase Corporation Engineering Dallas, Texas
kquick@iphase.com http://www.iphase.com 214.654.5173
-
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/

\
 
 \ /
  Last update: 2005-03-22 13:44    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site