Re: 2.4 kernel
> > That wasn't entirely clear from your mail. Anyway, the changes between 2.2
> > and 2.4 framebuffer drivers don't look to be much more than the rewritten
> > struct definitions. Backporting to 2.2.18 should be possible without major
> > hassle.
> Keep in mind that "major hassle" is relative.
Sure, I'm talking major hassle on my scale.
> Example: the first time I tried to do IP masquerading, it was a major hassle,
> because I started with the HOWTOs, and spent several hours trying to come up
> with good rules and ppp scripts. Then I searched the package archive in dselect
> and installing the ipmasq package did everything automatically, with a good set
> of conservative rules and the ppp scripts built-in, so it would now take me just
> four seconds (literally).
> It's like Miguel saying, "You want a new Gnumeric feature? Well, just add it to
> the source code!"
> It's not quite that simple for someone not familiar with the framebuffer
> sources. Maybe for the tdfx-fb author...
I'm not the tdfx author. I'm not even the macfb author. I've seen some
framebuffer code so I understand the basic design. My 'no major hassle'
call is based entirely on a diff between 2.2.18 and 2.4.0-test7 atyfb.c.
and what I've seen there was mostly cosmetics.
> In the meantime, I built -test6 successfully, and it boots with quik on a
> StarMax (pmac 4400 clone) and runs fine, except for ppp, but we've discussed
> that before. Couldn't get -test8 to build. Maybe -test6 will work with
Should work if the driver is in there?