Re: [Linux-fbdev-users] Framebuffer devices as modules?

[Geert U. <ge...@li...>]

On Sun, 28 Aug 2005, Sven Luther wrote:
> On Sun, Aug 28, 2005 at 12:16:54PM +0200, Geert Uytterhoeven wrote:
> > On Sat, 27 Aug 2005, Sven Luther wrote:
> > > On Sat, Aug 27, 2005 at 08:20:11AM +0800, Antonino A. Daplas wrote:
> > > > Sven Luther wrote:
> > > > >Yeah, well, the constraint is that this is on powerpc, and there is no 
> > > > >vgacon
> > > > >or whatever, and we still like to have debug output as soon as possible.
> > > > 
> > > > The framebuffer can load as early as you can load initrd. So you'll 
> > > > basically
> > > > miss bugs before the filesystem gets loaded. The screen is buffered, so the
> > > > contents are there. So if you are fortunate enough to load the framebuffer 
> > > > system
> > > > even if the bug happened a bit earlier, you might still be able to see the
> > > > message.
> > > > 
> > > > Note that the framebuffer system loads a little bit late compared to 2.4. 
> > > > It gets
> > > > loaded after PCI, PARISC and rapidio.
> > > 
> > > I guess we have always the serial console if earlier debugging is needed.
> > 
> > Don't we have promcon, on some PPCs?
> 
> Some, but not all. My main problem is that i build a generic powerpc kernel
> for debian, and right now we need many frambuffers builtin for the different
> machines. My idea is to build all those as modules, and put them in initramfs,
> and still call them early enough. Since initramfs, from what i am told, is
> accessed earlier than the initrd was, it will be more usable this way.

Another option would be to use offb, and find a way to `overtake' offb later by
a chipset-specific frame buffer device. A bit like vgacon is overtaken by fbdev
on other systems.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- ge...@li...

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds