Re: [PATCH] Open Firmware device tree virtual filesystem

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

> I'm incredibly surprised how much resistence there is from the
> i386 OFW folks to do this right.  It would be like 80 lines of
> code to suck the device tree into kernel memory, or if they don't
> want to do that they can use inline function wrappers to provide
> the clean C-language interface to all of this and the cost to
> i386-OFW would be zero with the benefit that other platforms could
> use the code potentially.
>
> Doing the same thing 3 different ways, knowingly, is just very bad
> engineering.  That is how you end up with a big fat pile of
> unmaintainable poo instead of a clean maintainable source tree.  If we
> fix a bug in one of these things, the other 2 are so different that if
> the bug is in the others we'll never know and it's not easy to check
> so people won't do it.
> 
> So please do this crap right.

I strongly agree. Nowadays, both powerpc and sparc use an in-memory copy
of the tree (wether you use the flattened format during the trampoline
from OF runtime to the kernel or not is a different matter, we created
that for the sake of kexec and embedded devices with no real OF, but the
end result is the same, a kernel based tree structure).

There is already powerpc's /proc/device-tree and sparc's openpromfs, I'm
all about converging that to a single implementation (a filesystem is
fine) that uses the in-memory tree.

Ben.