> On Wed, Feb 25, 2004 at 01:37:43PM +0100, Michael Schmitz wrote:
> > I've heard rumours about similar breakage from other users regarding
> > different hardware (including plain mach64). From what I gather, BenH
> > (or Linus) has started to incorporate changes from the fbdev devel tree. I
> > was hoping to keep out of this until the dust settles, and didn't dare
> > upgrade yet. It looks like you already checked for the obvious; I'd bite
> > the bullet and revert to the 2.6.2 EDID parsing code.
> fbdev.c changed a lot between 2.6.2 and 2.6.3. I'll give it a try if I
> find the time...
Yep, I now recall that setting CONFIG_FB_ATY_GENERIC_LCD on powerbooks is
horribly broken since it attempts to read the BIOS data. Building without
is fine, apparently (on the Lombard, at least). I wonder if substituting
the BIOS data check with data supplied by OF can help to make the generic
LCD code useful for Powerbooks.
Comparing the EDID code from 2.6.0-ben3 with the current code, I see two
differences: the old code used to look for OF properties named EDID,B and
EDID,A. And it would reject properties not of type "LCD". Does
changing these bits help?
Michael
