Subject: Re: Radeon framebuffer (the story continues)
To: Vincent <10.50@free.fr>
From: Michael Lorenz <macallan@netbsd.org>
List: tech-kern
Date: 12/28/2006 15:21:53
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Dec 28, 2006, at 04:58, Vincent wrote:

>>> None whatsoever. I guess nobody cares about a piece of silicium 5=20
>>> years old (read: heap of crap) that only knows about NetBSD! :)
>> Well, you obviously do and that's good enough I guess :p
>
> Remember: the needs of the many outweight the needs of the few... or=20=

> the one. ;)

Not necessarily in an all-volunteer project ;)
Most hardware I'm writing code for is more than 5 years old - I just=20
use / make work what I have here. If that benefits anyone else - fine=20
with me but I can't afford getting more modern hardware just because=20
more people might use something similar.

>> Does Linux do The Right Thing on your laptop? Maybe they have=20
>> something readable in the kernel.
>
> I've tried to find the Linux' radeonfb.c file. If I have stumbled upon=20=

> the right one, it doesn't seem to read anything from the BIOS, just=20
> defaulting to typical cases. It is of no real use. But I may have=20
> erred.

I think it's in linux/drivers/video/ati or something like that. What I=20=

meant was - did you try to boot some recent linux CD which uses an fb=20
console ( or draws a splashscreen or something ) ? Maybe they just=20
guess better than radeonfb, if they do we should probably follow.

>>> I don't like to cut/copy verbatim pieces of software that I don't=20
>>> understand.
>> Nobody asked you to.
>
> No, that's right. But eventually, if you don't own any other reliable=20=

> source, you have to. At least you have to copy it to some degree. That=20=

> means two things: first, you have to trust the code you imitate, and,=20=

> if there is a bug, you can't even figure it out and fix it; second, I=20=

> am unsure whether such a code can be licenced BSD if it is drawn out a=20=

> GPL or LGPL'ed software.

Yeah, that's why I usually avoid copying anything from linux, just read=20=

it for the comments and #defines, then write my own. Besides that they=20=

tend to use structs for register layouts which is totally un-NetBSD=20
since it makes it impossible or dead ugly to use the bus_space_* stuff.

>> Because ATI seems to think those are trade secrets or something.
>> They apparently think that giving away the specs means to allow=20
>> others to copy their designs.
>> I really don't understand why ATI doesn't even publish specs for=20
>> older chips ( like all the mach64 variants, rage128 or older radeons=20=

>> ) - even Intel does that, docs for many C&T chips are freely=20
>> available from intel.com.
>
> Well, Intel=AE seems to have a natural position: the best you know a=20=

> circuit, the more likely you are to use it.

Yeah, an operating system agnostic view is what I'd expect from a=20
hardware firm - what would they care /who/ uses their chips as long as=20=

they get them sold?

have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Darwin)

iQEVAwUBRZQnYcpnzkX8Yg2nAQLFZAf9GKL8FsUEp8Hv8eux42PtF+TXBgbU1G9H
2c4jChc/AQ06olPJ6dCrGVleRoaARA0WvlM72s3LgJjXMXcP0+QBF4zDSNuXLKUL
hoiZOJT2EImB0yjc3u4+uJOBjSKEa7rOSoTq9gAxAGKut8YM8lLxh7d+IJTPi5F9
Qj55fwqSgjUZf9xz44D+6TRDjw+yG1Lf0HFooE84n6b/cSkhaFrj+Fh7R9bvV+bX
RrnWhUhEGjAwj2tgkNxQJ2ZXC6oRHrzmv1kegI0anww3yn2CH9G2RATAhGY5PzBC
S8lqI+h6TD6bSVp5SITI6XXl7p8eDmhomhQww7qsawDbU0nsvm720w=3D=3D
=3D417r
-----END PGP SIGNATURE-----