Re: PnP support?

[Randy Heineke <heineke@vanilla.r.umn.edu>]

> People were arguing back and forth earlier about whether our install
> system should use PnP detection or not. I don't know much about the
> current hardware detection/probing that we're planning; I do know that
> I want the seul installer to figure out as much as possible about the
> hardware of the system it's on.

I absolutely agree with you.  If the only users who can use Linux are the
ones that have in-depth knowledge of there devices SEUL is just about
pointless.  

I do not wish to get into a technical debate with Linus Torvalds, I have
nothing but admiration for his work.  I do find his response reasonable
to the degree that I understand it.  I think he is saying that PnP does
not belong in the Linux kernel.  There are many things that I do not
understand.  Some of these we can sort out among ourselves.  We may wish
to come back to Linus with a few questions though. Here is what I do not
understand.

1. PnP does not belong in the kernel but is there a place for a kernel
mode that allows addressing in protected ranges early in the start up
process that gives us a chance to interegate hardware? 

2. Does anyone know what mechanisms cause Linux to automatically detect
hardware cahnges?

3. Does anyone know what Alan is up to?
 
> Does anybody have actual facts on PnP?
> 
> --Roger
> 
> (post from linux-kernel)
> 
> From: Linus Torvalds <torvalds@transmeta.com>
> Date: Fri, 3 Apr 1998 18:29:28 -0800 (PST)
> Subject: Re: Linux-2.1.92 - Feature Freeze 
> 
> On Fri, 3 Apr 1998, Horst von Brand wrote:
> > 
> > > No, but Alan is looking into some minimal PnP support. I discussed this
> > > at some length with him, and the way I would have preferred it done he
> > > didn't think we could do in time for 2.2, so we'll have some very simple
> > > groundwork but not a "real" PnP thing.
> > 
> > Is this really wise? The moment 2.2 comes out, people will be screaming for
> > PnP support, and it will have to be integrated into the _stable_ kernel.
> > And there goes my wish for 2.2: 2.2.4 at most ;-)
> 
> It really is wise.
> 
> The MicroSoft PnP standard is truly a really _bad_ standard. I refuse to
> have that kind of setup in my kernel - the whole point with Linux it to
> have an operating system that is _better_ than the crud MS keeps throwing
> out. 
> 
> Have you noticed how well Linux autodetects hardware even _without_ any PR
> sticker that says "PnP" on it? PnP is just a marketing term. We need some
> very minimal support to avoid trampling on some IO ranges etc, but we
> don't want anything more than the minimal requirements, because we aren't
> going to make the same silly mistakes MS does. 
> 
>                 Linus
> 
>