[coreboot] [PATCH] First step in converting W83977TF early serial from included to linked

[Keith Hui]

And here is the patch. abuild-tested. I will boot test it with P2B-LS
and P3B-F tomorrow but I want this patch out there to generate some
discussions and get more boot test coverage.

This I believe falls under "infrastructure projects" [1].

This patch facilitates, for boards using Winbond W83977TF superio,
dropping early_serial.c from #includes of their romstage, instead
linking to them; and converts 12 of 13 mainboards using this superio
to do exactly this. The lone board out, iei/nova4899r, is a
Geode/CS5530 board that has not yet been converted to CAR or tiny
bootblock. The rest are all slot 1/440BX/i82371eb boards that have
been converted. At this stage I should leave converting CS5530 to tiny
bootblock to someone better versed in this platform.

The pnp_... functions defined in romcc_io.h now have a new home in
arch/x86/lib/pnp.c, and is compiled and linked like any other C files
meant for romstage.

This patch puts the W83977TF early serial code in a state where it can
be incorporated both through the old setup (ie. #included in mainboard
romstage), or be compiled separately and linked into romstage. Once
this conversion is done on all superios and all southbridges/boards
have been converted, the few #ifdefs that made this possible can be
cleaned out.

Signed-off-by: Keith Hui

[1] http://www.coreboot.org/Infrastructure_Projects#Remove_.c_includes

On Mon, Feb 28, 2011 at 1:00 PM, Keith Hui  wrote:
> Hi all,
>
> In my pipeline is something more infrastructural, something different
> from what I usually do -
>
> I am converting early serial code for Winbond W83977TF from code
> included in romstage to compiled unit to be linked into romstage.
>
> 13 boards use this superio, some apparently still not converted to CAR
> (including one Intel server board). I have yet to complete running
> abuild on my change. Should I let it break when I submit it for review
> given earlier buzz suggesting we leave non-CAR boards behind?
>
> The reason I'm doing this is I want to do this for ITE IT8705F, which
> happens to have no board using it at the moment, but I want to test
> this approach on a board that I can boot test right now. Enter the
> P2B-LS and W83977TF. If this proof of concept works, we can start this
> conversion one superio at a time and nail down one more goal of the
> infrastructure refactoring.
>
> Thoughts and comments?

Thanks
Keith