Re: Where arch/bebox/pci/pfb.c?

[Michael Lorenz <macallan%netbsd.org@localhost>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On May 17, 2008, at 12:46, KIYOHARA Takashi wrote:

> From: KIYOHARA Takashi <kiyohara%kk.iij4u.or.jp@localhost>
> Date: Thu, 15 May 2008 15:54:15 +0900 (JST)
>
> > > I think what the correct thing to do in your case, would be to  
> > > start with
> > > genfb, and get that working.  In theory, it should work with any  
> > > card, as long
> > > as you know the current display mode of the card.  Once you have  
> > > a working
> > > console with genfb, then maybe look into writing a proper  
> > > matroxfb or s3virgefb.
> >
> > hmm... I can't understand. X-<
> > Should I attach genfb?  In that case, how should be done the  
> > console?  Do
> > not we have a console until genfb is attached?
> >
> > Shall I define?
> >
> >   # generic PCI framebuffers
> >   genfb*          at pci?
> >
> >   # make sure the console display is always wsdisplay0
> >   wsdisplay0      at wsemuldisplaydev? console 1
> >   wsdisplay*      at wsemuldisplaydev?
>
> I succeed to support genfb(4).  ;-)

Nice :)

> However, I saw the color of the character and the cursor strange.  For
> instance, the cursor was a pink color, font was a gray.  I was able to
> change the color to white by defining WS_DEFAULT_FG.
>
>     options WS_DEFAULT_FG=WSCOL_LIGHT_WHITE

Genfb doesn't know anything about the underlying hardware, to access  
colour map registers it needs a callback ( see spar64/sparc64/ 
autoconf.c for example, or powerpc/oea/ofw_autoconf.c )
If the firmware sets up a sane colour map I'll just add support for a  
static colour map, rasops can already deal with it.

> Moreover, I was able to obtain following information.
>
>     0x00 - 0x1f : black - white
>     0x20 - 0x3f :       - blue
>     0x40 - 0x5f :       - red
>
>
> And I have a new problem.  Both of the match() of vga and genfb  
> returns 1.
> Therefore, which attaching is tried cannot be specified.  Isn't it  
> return
> value revokable dynamic in some methods we?

genfb is intended to be a catch-all fallback for graphics cards where  
we can find the framebuffer but don't have a native driver for it.  
Drivers like chipsfb, machfb etc. all return 100 they match something  
so they'd win against genfb.

If vga works I'd say stick with it.

have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBSC8bEspnzkX8Yg2nAQJpwAgAsQbs1q6OBv5z+eIt5OG4Q1pPNvbm1JDw
oIfxakpNyLvkisJbMMXuJ7rf9cyMWQpbagOriltftaX1gcCIEyFvGvFO/LDP4vNg
88ecunfaOKMb97xZMbmUVpQ9Yz81MQZeFJhZxLbuygmPsddmCbuLGSBkwhSHbVW7
vx35jsBa7g5per1BdZnjRYxV13m0lcpNOCTPv1BO6tSDopshDYVukfXFWn5jNlWQ
yRKai/5DpQggAlhUEfWnOrvqqkjWgOTwETVO3FCC6F73w9gRz2IMVOmnUGlFVkpL
CwIg7v+Zh1ypZvcltIrrHb+oJD7vC3b3BCzgzO/iv4AWgXUvCpENug==
=HsLG
-----END PGP SIGNATURE-----