On Mon, 10 Jan 2005 08:19:09 -0800
Chuck Silvers <chuq@chuq.com> wrote:

> 
> this seems pretty reasonable.  I cleaned up the patch (and removed the
> font-size fiddling, since it's unrelated), it's attached.

I wouldn't commit this just yet. Brian Hechinger and I are working on
his Radeon card that appears to be the exception that proves the rule.
After he flashed it for MacOS, it is now properly being seen in OF and
has some quirks. The Radeon card itself is not identifying itself in OF
as a "display" device, although it does carry the PCI class of display.
Instead, it has two child nodes that are both OF frame buffer compliant,
or at least appear to be, as we're having some trouble getting OF to
actually use them. If we can get OF to use either one, I'm not sure if
the best way to get NetBSD to see them is to iterate through the device
if it carries PCI_CLASS_DISPLAY and look for something OF compliant, or
to write a skeleton driver to attach the Radeon card and then subattach
the children nodes, while adding something like ofb at pci, rad to
files.macppc and a rad* at pci device.

tim