[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Wireless PCI card support in upcoming 3.6



On Tuesday 14 September 2004 22.16, Peter H. Coffin wrote:
> On Tue, Sep 14, 2004 at 09:37:27PM +0200, Per-Olov Sjöholm wrote:
> > On Tuesday 14 September 2004 21.09, you [Theo] wrote:
> > > If you need 802.11g support you will find that it is not in OpenBSD.
> > >
> > > There are no free drivers, and we strongly believe in free software.
> > >
> > > If this is a problem, we urge you to contact the vendors.
> >
> > Sorry Theo if you receive this twice (forgot the list in the reply...)
> >
> > Hmmm...ok
> >
> > Why isn't it a problem with wireless cards for lowers speeds as well
> > ? I have not seen one driver ever for OpenBSD from any vendor. But
> > still, we have alot of drivers ... So I do not really understand. Can
> > somebody then explain how driver development works. Are all drivers
> > ports from NetBSD ? Or are all the drivers written by the OpenBSD team
> > ? If so... what is the problem in writing drivers for the 802.11g
> > cards ? Wont the vendors give out specs ? If so, how can drivers for
> > linux exist for the cards ? Or is it the very free "BSD" license that
> > causes this ?
> >
> > Can anybody give a decent explanation to a person that do not
> > understand the driver development to 100% (Or is it maybe the BSD
> > license I have to read again). I think more people than me should be
> > interested in this...
>
> Roughly...
>
> Some cards running at lower speeds are using chips that have published
> documentation of how write software that makes their cards go, without
> restrictions on how that documentation is used. The references to "Prism
> chips" are a common one. Since the information about how to make the
> Prism II chips go is available, software ("drivers") is created by
> people that believe in licensing software in a manner that's compatible
> with the BSD license, and that software can be incorporated into
> OpenBSD.
>
> Other chip manufacturers feel that they are the only ones that should be
> allowed to write software that works with their chips. They give the
> card manufacturers the software on very restrictive terms that might
> (for example) say "Thou shalt only distribute this software with an
> accompanying product that contains our chips." These products usually
> end up with only Windows logos on the box, and you do not find drivers
> for MacOS or Linux or anything else, ever.
>
> A third sort will allow others to write software, but only will pass out
> documentation to developers under non-disclosure agreements (NDAs). The
> developers cannot tell anyone else how the chips work, and must make a
> reasonable effort to make sure that other clever developers that have
> NOT become party to the NDAs don't figure out how to make the chips
> work without becoming equally part of the NDA. These kind of chips end
> up with "binary-only" drivers. The drivers are redistributable and
> can be bundled with software releases but cannot be reveiwed, changed
> or updated except by the person or group that wrote them in the first
> place. You may see drivers for OSX or linux, or FreeBSD, but you won't
> see them for OpenBSD.
>
> At this point, NONE of the makers of 802.11g chips have released
> documentation for how to make their chips go in the first manner. If
> that ever happens, then you might see OpenBSD support 802.11g cards.
> Until it does, you won't.

Thanks for the very good reply that answer a lot of my questions.


t seems like the thread "WIFI: Conexant PRISM ?" also discuss driver issues 
acting like they were 10 year old yelling at each other. Don't understand why 
peope can't be polite even if they are upset. But the driver debate seems to 
be hot right now...

So it seems like OpenBSD will die at the end if all vendors start to request 
NDA:s for driver development which will by no means be accepted by OpenBSD... 
Hopefully it will be a solution to the problems before my favorite OS will 
die a driverless death. 


Are there no ways around the problem? What is the real problem with for 
example:
* A wrapper like Anthony Gabrielson mention in this thread (last choice).
* kernel loadable drivers outside the OS distribution. 
* Use drivers written under NDA as FreeBSD.  Is it security only thoughts that 
there is a "no binary only code" acceptance?
If I remeber it right there where earlier a hot debate regarding kernel 
loadable modules as well. But I do not remember how that debate ended.


Some clever person at the core OpenBSD development team must have had the 
thought:  "Hmmm what happens with OpenBSD in the future if more and more  
vendors start to require NDA:s? What do we do then.." 

I am very interesting to know what the core developers (and Theo as well) 
really think as I am very interested in OpenBSD:s survival.  And I agree with 
Ed White in the other  "WIFI: Conexant PRISM ?" thread that it would at least 
be very nice to know the future plans... I relly hope I do not have to in the 
future switch top FreeBSD to be able to use a standard market leader card 
rejected by OpenBSD caused by an NDA agreement requirement for driver 
development.

And I appreciate if we (all that read this) can avoid short childish answers 
like "It will never happen". This as we all hopefully want to use the great 
OpenBSD in the future as well. 

Thanks
/Per-Olov