The reason behind this, is that previously (i.e. 2.4) agp_init()
would call the chipset specific backend init routines.
In 2.5, that stuff got pushed out into seperate modules,
so now i810fb does a (superflous) agp_init() which just
printk's the agpgart banner, and then calls the intel_agp_init()
function directly.
This *should* work, but doesn't.
The agpgart code *absolutely must* be set up before the framebuffer
code. Previously this was going to be solved with link order shuffling,
however that won't work, as the drivers/video stuff is initialised
from drivers/char/mem.c (Really non-intuitive place to hide fb init IMO)
mem.c gets linked before agp/ so we lose there.
Hence the explicit init in i810fb. I've tried to untangle this and
figure out why that isn't working, at a complete loss.
(The lack of hardware to test this on also doesn't help).
Any ideas ?
Dave
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/