matthew green <mrg@eterna.com.au> writes:
> the one remaining problem is getting the OFW node for children that need
> it, such as the ebus on sparc64.  i'd really like to get rid of the hack
> that finds the `ebus' node from that driver...

If you find that you have a general problem like that, it may well
make sense to define a machine-dependent (or even common to OFW-using
ports) function which translates from PCI chipset + tag -> ofw node.

That should be reasonable to do -- but probably no less of a hack (to
some eyes 8-) -- than the code in ebus.c.


maybe something like "pci_find_ofw_node(chipset, tag)".

That's about the best I can suggest.  Given that it's MD, though, it
kinda does make sense... c.f. the 'decompose tag' fns used by some
ports.  This is, in some ways, a more complex version of one of
them. 8-)

(of course, it looks like people have started to use
pci_decompose_tag() in MI code without actually defining that as an MI
API... but that's another problem.  8-)



cgd