Re: [coreboot] Native init for AMD cards

[ron minnich]

This is the sort of thing I started with. It looks easier at first,
but as you hit kernel churn, white space issues, the need to delete
struct members or even rewrite structs and the code that uses them,
and a lot of other trivial things, I expect you to quickly find you
need a semantic patch. But the proof is in the pudding -- just give it
a try and see how it goes.

Note that on i915tool, I got about 50% of the way there with
coccinnelle, and even had things working in user mode, but my new
approach is to build up code from the traced IO stream. And, at the
end, there's no way to avoid a lot of hand work, it seems. But maybe
the intel stuff is particularly hard, that seems to be the rule.

ron